Association of Heart Failure With Cognitive Decline and Development of Mild Cognitive Impairment and Dementia.

BACKGROUND: Incidence of cognitive impairment and its consequences have not been fully examined in heart failure (HF). OBJECTIVE: The aim of this study was to examine associations of HF with cognitive decline, frequencies and risks of, and time-to-develop mild cognitive impairment (MCI) or dementia during 15-year follow-up. METHODS: For this retrospective cohort study, data were retrieved from the National Alzheimer's Coordinating Center. Cognitive decline was assessed using the Uniform Data Set neuropsychological battery. Development of MCI and dementia was assessed using clinically diagnosed cognitive status. RESULTS: Compared with participants without HF (n = 12 904), participants with HF (n = 256) had more decline in attention, executive function, and memory while controlling for covariates including apolipoprotein E4. Participants with HF developed MCI or dementia more frequently (44.9% vs 34.4%), developed dementia faster from normal cognition, and had a lower risk of dementia from MCI after controlling for covariates (hazard ratio, 0.71) than participants without HF. CONCLUSIONS: Heart failure was associated with accelerated cognitive decline.

Integrated Single-Cell Multiomic Profiling of Caudate Nucleus Suggests Key Mechanisms in Alcohol Use Disorder.

Alcohol use disorder (AUD) is likely associated with complex transcriptional alterations in addiction-relevant brain regions. We characterize AUD-associated differences in cell type-specific gene expression and chromatin accessibility in the caudate nucleus by conducting a single-nucleus RNA-seq assay and a single-nucleus RNA-seq + ATAC-seq (multiome) assay on caudate tissue from 143 human postmortem brains (74 with AUD). We identified 17 cell types. AUD was associated with a higher proportion of microglia in an activated state and more astrocytes in a reactive state. There was widespread evidence for differentially expressed genes across cell types with the most identified in oligodendrocytes and astrocytes, including genes involved in immune response and synaptic regulation, many of which appeared to be regulated in part by JUND and OLIG2. Microglia-astrocyte communication via interleukin-1 beta, and microglia-astrocyte-oligodendrocyte interaction via transforming growth factor beta 1 were increased in individuals with AUD. Expression quantitative trait loci analysis revealed potential driver genes of AUD, including ADAL, that may protect against AUD in medium spiny neurons and interneurons. This work provides a thorough profile of the effects of AUD in the human brain and identifies several promising genes for further study.

Genome-wide meta-analyses of cross substance use disorders in European, African, and Latino ancestry populations.

Genetic risks for substance use disorders (SUDs) are due to both SUD-specific and SUD-shared genes. We performed the largest multivariate analyses to date to search for SUD-shared genes using samples of European (EA), African (AA), and Latino (LA) ancestries. By focusing on variants having cross-SUD and cross-ancestry concordant effects, we identified 45 loci. Through gene-based analyses, gene mapping, and gene prioritization, we identified 250 SUD-shared genes. These genes are highly expressed in amygdala, cortex, hippocampus, hypothalamus, and thalamus, primarily in neuronal cells. Cross-SUD concordant variants explained ~ 50% of the heritability of each SUD in EA. The top 5% individuals having the highest polygenic scores were approximately twice as likely to have SUDs as others in EA and LA. Polygenic scores had higher predictability in females than in males in EA. Using real-world data, we identified five drugs targeting identified SUD-shared genes that may be repurposed to treat SUDs.

Unsupervised representation learning on high-dimensional clinical data improves genomic discovery and prediction.

Although high-dimensional clinical data (HDCD) are increasingly available in biobank-scale datasets, their use for genetic discovery remains challenging. Here we introduce an unsupervised deep learning model, Representation Learning for Genetic Discovery on Low-Dimensional Embeddings (REGLE), for discovering associations between genetic variants and HDCD. REGLE leverages variational autoencoders to compute nonlinear disentangled embeddings of HDCD, which become the inputs to genome-wide association studies (GWAS). REGLE can uncover features not captured by existing expert-defined features and enables the creation of accurate disease-specific polygenic risk scores (PRSs) in datasets with very few labeled data. We apply REGLE to perform GWAS on respiratory and circulatory HDCD-spirograms measuring lung function and photoplethysmograms measuring blood volume changes. REGLE replicates known loci while identifying others not previously detected. REGLE are predictive of overall survival, and PRSs constructed from REGLE loci improve disease prediction across multiple biobanks. Overall, REGLE contain clinically relevant information beyond that captured by existing expert-defined features, leading to improved genetic discovery and disease prediction.

Analysis of the combined effect of rs699 and rs5051 on angiotensinogen expression and hypertension.

Background: Hypertension (HTN) involves genetic variability in the renin-angiotensin system and influences antihypertensive response. We previously reported that angiotensinogen (AGT) messenger RNA (mRNA) is endogenously bound by miR-122-5p and rs699 A > G decreases reporter mRNA in the microRNA functional-assay PASSPORT-seq. The AGT promoter variant rs5051 C > T is in linkage disequilibrium (LD) with rs699 A > G and increases AGT transcription. The independent effect of these variants is understudied due to their LD therefore we aimed to test the hypothesis that increased AGT by rs5051 C > T counterbalances AGT decreased by rs699 A > G, and when these variants occur independently, it translates to HTN-related phenotypes. Methods: We used in silico, in vitro, in vivo, and retrospective models to test this hypothesis. Results: In silico, rs699 A > G is predicted to increase miR-122-5p binding affinity by 3%. Mir-eCLIP results show rs699 is 40-45 nucleotides from the strongest microRNA-binding site in the AGT mRNA. Unexpectedly, rs699 A > G increases AGT mRNA in an AGT-plasmid-cDNA HepG2 expression model. Genotype-Tissue Expression (GTEx) and UK Biobank analyses demonstrate liver AGT expression and HTN phenotypes are not different when rs699 A > G occurs independently from rs5051 C > T. However, GTEx and the in vitro experiments suggest rs699 A > G confers cell-type-specific effects on AGT mRNA abundance, and suggest paracrine renal renin-angiotensin-system perturbations could mediate the rs699 A > G associations with HTN. Conclusions: We found that rs5051 C > T and rs699 A > G significantly associate with systolic blood pressure in Black participants in the UK Biobank, demonstrating a fourfold larger effect than in White participants. Further studies are warranted to determine if altered antihypertensive response in Black individuals might be due to rs5051 C > T or rs699 A > G. Studies like this will help clinicians move beyond the use of race as a surrogate for genotype.

Systematic rare variant analyses identify RAB32 as a susceptibility gene for familial Parkinson's disease.

Despite substantial progress, causal variants are identified only for a minority of familial Parkinson's disease (PD) cases, leaving high-risk pathogenic variants unidentified1,2. To identify such variants, we uniformly processed exome sequencing data of 2,184 index familial PD cases and 69,775 controls. Exome-wide analyses converged on RAB32 as a novel PD gene identifying c.213C > G/p.S71R as a high-risk variant presenting in ~0.7% of familial PD cases while observed in only 0.004% of controls (odds ratio of 65.5). This variant was confirmed in all cases via Sanger sequencing and segregated with PD in three families. RAB32 encodes a small GTPase known to interact with LRRK2 (refs. 3,4). Functional analyses showed that RAB32 S71R increases LRRK2 kinase activity, as indicated by increased autophosphorylation of LRRK2 S1292. Here our results implicate mutant RAB32 in a key pathological mechanism in PD-LRRK2 kinase activity5-7-and thus provide novel insights into the mechanistic connections between RAB family biology, LRRK2 and PD risk.

Decoding the Role of Secondary Motor Cortex Neuronal Ensembles during Cocaine Self-Administration: Insights from Longitudinal in vivo Calcium Imaging via Miniscopes.

Recent findings in our lab demonstrated that the risk of cocaine relapse is closely linked to the hyperexcitability of cortical pyramidal neurons in the secondary motor cortex (M2), noticeable 45 days after cocaine intravenous self-administration (IVSA). The present study was designed to explore the underlying mechanisms of neuronal alterations in M2. Our hypothesis was that M2 neurons were affected directly by cocaine taking behaviors. This hypothesis was tested by monitoring individual neuronal activity in M2 using MiniScopes for in vivo Ca 2+ imaging in C57BL/6J mice when they had access to cocaine IVSA as a reinforcement (RNF) contingent to active lever press (ALP) but not to inactive lever press (ILP). With support of our established pipeline to processing Ca 2+ imaging data, the current study was designed to monitor M2 neuronal ensembles at the single-neuron level in real time with high temporal resolution and high throughput in each IVSA session and longitudinally among multiple IVSA sessions. Specifically, five consecutive 1-hr daily IVSA sessions were used to model the initial cocaine taking behaviors. Besides detailed analyses of IVSA events (ALP, ILP, and RNF), the data from Ca 2+ imaging recordings in M2 were analyzed by (1) comparing neuronal activation within a daily IVSA session (i.e., the first vs. the last 15 min) and between different daily sessions (i.e., the first vs. the last IVSA day), (2) associating Ca 2+ transients with individual IVSA events, and (3) correlating Ca 2+ transients with the cumulative effects of IVSA events. Our data demonstrated that M2 neurons are exquisitely sensitive to and significantly affected by concurrent operant behaviors and the history of drug exposure, which in turn sculpt the upcoming operant behaviors and the response to drugs. As critical nodes of the reward loop, M2 neurons appear to be the governing center orchestrating the establishment of addiction-like behaviors.

Utilizing multimodal AI to improve genetic analyses of cardiovascular traits.

Electronic health records, biobanks, and wearable biosensors contain multiple high-dimensional clinical data (HDCD) modalities (e.g., ECG, Photoplethysmography (PPG), and MRI) for each individual. Access to multimodal HDCD provides a unique opportunity for genetic studies of complex traits because different modalities relevant to a single physiological system (e.g., circulatory system) encode complementary and overlapping information. We propose a novel multimodal deep learning method, M-REGLE, for discovering genetic associations from a joint representation of multiple complementary HDCD modalities. We showcase the effectiveness of this model by applying it to several cardiovascular modalities. M-REGLE jointly learns a lower representation (i.e., latent factors) of multimodal HDCD using a convolutional variational autoencoder, performs genome wide association studies (GWAS) on each latent factor, then combines the results to study the genetics of the underlying system. To validate the advantages of M-REGLE and multimodal learning, we apply it to common cardiovascular modalities (PPG and ECG), and compare its results to unimodal learning methods in which representations are learned from each data modality separately, but the downstream genetic analyses are performed on the combined unimodal representations. M-REGLE identifies 19.3% more loci on the 12-lead ECG dataset, 13.0% more loci on the ECG lead I + PPG dataset, and its genetic risk score significantly outperforms the unimodal risk score at predicting cardiac phenotypes, such as atrial fibrillation (Afib), in multiple biobanks.

Functional Analysis of G6PD Variants Associated With Low G6PD Activity in the All of Us Research Program.

Glucose-6-phosphate dehydrogenase (G6PD) protects red blood cells against oxidative damage through regeneration of NADPH. Individuals with G6PD polymorphisms (variants) that produce an impaired G6PD enzyme are usually asymptomatic, but at risk of hemolytic anemia from oxidative stressors, including certain drugs and foods. Prevention of G6PD deficiency-related hemolytic anemia is achievable through G6PD genetic testing or whole-genome sequencing (WGS) to identify affected individuals who should avoid hemolytic triggers. However, accurately predicting the clinical consequence of G6PD variants is limited by over 800 G6PD variants which remain of uncertain significance. There also remains significant variability in which deficiency-causing variants are included in pharmacogenomic testing arrays across institutions: many panels only include c.202G>A, even though dozens of other variants can also cause G6PD deficiency. Here, we seek to improve G6PD genotype interpretation using data available in the All of Us Research Program and using a yeast functional assay. We confirm that G6PD coding variants are the main contributor to decreased G6PD activity, and that 13% of individuals in the All of Us data with deficiency-causing variants would be missed if only the c.202G>A variant were tested for. We expand clinical interpretation for G6PD variants of uncertain significance; reporting that c.595A>G, known as G6PD Dagua or G6PD Açores, and the newly identified variant c.430C>G, reduce activity sufficiently to lead to G6PD deficiency. We also provide evidence that five missense variants of uncertain significance are unlikely to lead to G6PD deficiency, since they were seen in hemi- or homozygous individuals without a reduction in G6PD activity. We also applied the new WHO guidelines and were able to classify two synonymous variants as WHO class C. We anticipate these results will improve the accuracy, and prompt increased use, of G6PD genetic tests through a more complete clinical interpretation of G6PD variants. As the All of Us data increases from 245,000 to 1 million participants, and additional functional assays are carried out, we expect this research to serve as a template to enable complete characterization of G6PD deficiency genotypes. With an increased number of interpreted variants, genetic testing of G6PD will be more informative for preemptively identifying individuals at risk for drug- or food-induced hemolytic anemia.

Functional 3'-UTR Variants Identify Regulatory Mechanisms Impacting Alcohol Use Disorder and Related Traits.

Although genome-wide association studies (GWAS) have identified loci associated with alcohol consumption and alcohol use disorder (AUD), they do not identify which variants are functional. To approach this, we evaluated the impact of variants in 3' untranslated regions (3'-UTRs) of genes in loci associated with substance use and neurological disorders using a massively parallel reporter assay (MPRA) in neuroblastoma and microglia cells. Functionally impactful variants explained a higher proportion of heritability of alcohol traits than non-functional variants. We identified genes whose 3'UTR activities are associated with AUD and alcohol consumption by combining variant effects from MPRA with GWAS results. We examined their effects by evaluating gene expression after CRISPR inhibition of neuronal cells and stratifying brain tissue samples by MPRA-derived 3'-UTR activity. A pathway analysis of differentially expressed genes identified inflammation response pathways. These analyses suggest that variation in response to inflammation contributes to the propensity to increase alcohol consumption.

Clinical, genomic, and neurophysiological correlates of lifetime suicide attempts among individuals with alcohol dependence.

Research has identified clinical, genomic, and neurophysiological markers associated with suicide attempts (SA) among individuals with psychiatric illness. However, there is limited research among those with an alcohol use disorder (AUD), despite their disproportionately higher rates of SA. We examined lifetime SA in 4,068 individuals with DSM-IV alcohol dependence from the Collaborative Study on the Genetics of Alcoholism (23% lifetime suicide attempt; 53% female; 17% Admixed African American ancestries; mean age: 38). We 1) conducted a genome-wide association study (GWAS) of SA and performed downstream analyses to determine whether we could identify specific biological pathways of risk, and 2) explored risk in aggregate across other clinical conditions, polygenic scores (PGS) for comorbid psychiatric problems, and neurocognitive functioning between those with AD who have and have not reported a lifetime suicide attempt. The GWAS and downstream analyses did not produce any significant associations. Participants with an AUD who had attempted suicide had greater rates of trauma exposure, major depressive disorder, post-traumatic stress disorder, and other substance use disorders compared to those who had not attempted suicide. Polygenic scores for suicide attempt, depression, and PTSD were associated with reporting a suicide attempt (ORs = 1.22-1.44). Participants who reported a SA also had decreased right hemispheric frontal-parietal theta and decreased interhemispheric temporal-parietal alpha electroencephalogram resting-state coherences relative to those who did not, but differences were small. Overall, individuals with alcohol dependence who report SA appear to experience a variety of severe comorbidities and elevated polygenic risk for SA. Our results demonstrate the need to further investigate suicide attempts in the presence of substance use disorders.

Associations between alcohol use disorder polygenic score and remission in participants from high-risk families and the Indiana Biobank.

BACKGROUND: In the United States, ~50% of individuals who meet criteria for alcohol use disorder (AUD) during their lifetimes do not remit. We previously reported that a polygenic score for AUD (PGSAUD ) was positively associated with AUD severity as measured by DSM-5 lifetime criterion count, and AUD severity was negatively associated with remission. Thus, we hypothesized that PGSAUD would be negatively associated with remission. METHODS: Individuals of European (EA) and African ancestry (AA) from the Collaborative Study on the Genetics of Alcoholism (COGA) who met lifetime criteria for AUD, and two EA cohorts ascertained for studies of liver diseases and substance use disorders from the Indiana Biobank were included. In COGA, 12-month remission was defined as any period of ≥12 consecutive months without meeting AUD criteria except craving and was further categorized as abstinent and non-abstinent. In the Indiana Biobank, remission was defined based on ICD codes and could not be further distinguished as abstinent or non-abstinent. Sex and age were included as covariates. COGA analyses included additional adjustment for AUD severity, family history of remission, and AUD treatment history. RESULTS: In COGA EA, PGSAUD was negatively associated with 12-month and non-abstinent remission (p ≤ 0.013, ßs between -0.15 and -0.10) after adjusting for all covariates. In contrast to the COGA findings, PGSAUD was positively associated with remission (p = 0.004, ß = 0.28) in the Indiana Biobank liver diseases cohort but not in the Indiana Biobank substance use disorder cohort (p = 0.17, ß = 0.15). CONCLUSIONS: PGSAUD was negatively associated with 12-month and non-abstinent remission in COGA EA, independent of behavioral measures of AUD severity and family history of remission. The discrepant results in COGA and the Indiana Biobank could reflect different ascertainment strategies: the Indiana Biobank participants were older and had higher rates of liver disease, suggesting that these individuals remitted due to alcohol-related health conditions that manifested in later life.

Polygenic Contributions to Suicidal Thoughts and Behaviors in a Sample Ascertained for Alcohol Use Disorders.

Introduction: Suicidal thoughts and behaviors have partially distinct genetic etiologies. Methods: We used PRS-CS to create polygenic risk scores (PRSs) from GWAS of non-suicidal self-injury, broad-sense self-harm ideation, nonfatal suicide attempt, death by suicide, and depression. Using mixed-effect models, we estimated whether these PRSs were associated with a range of suicidal thoughts and behaviors in the Collaborative Study on the Genetics of Alcoholism (N = 7,526). Results: All PRSs were significantly associated with suicidal ideation and suicide attempt (betas = 0.08-0.44, false discovery rate [FDR] <0.023). All PRSs except non-suicidal self-injury PRS were associated with active suicidal ideation (betas = 0.14-0.22, FDR <0.003). Several associations remained significant in models where all significant PRSs were included as simultaneous predictors, and when all PRSs predicted suicide attempt, the PRS together explained 6.2% of the variance in suicide attempt. Significant associations were also observed between some PRSs and persistent suicidal ideation, non-suicidal self-injury, compounded suicide attempt, and desire to die. Conclusion: Our findings suggest that PRS for depression does not explain the entirety of the variance in suicidal thoughts and behaviors, with PRS specifically for suicidal thoughts and behaviors making additional and sometimes unique contributions.

Identifying Genes Associated with Alzheimer's Disease Using Gene-Based Polygenic Risk Score.

BACKGROUND: Except APOE, Alzheimer's disease (AD) associated genes identified in recent large-scale genome-wide association studies (GWAS) had small effects and explained a small portion of heritability. Many AD-associated genes have even smaller effects thereby sub-threshold p-values in large-scale GWAS and remain to be identified. For some AD-associated genes, drug targeting them may have limited efficacies due to their small effect sizes. OBJECTIVE: The purpose of this study is to identify AD-associated genes with sub-threshold p-values and prioritize drugs targeting AD-associated genes that have large efficacies. METHODS: We developed a gene-based polygenic risk score (PRS) to identify AD genes. It was calculated using SNPs located within genes and having the same directions of effects in different study cohorts to exclude cohort-specific findings and false positives. Gene co-expression modules and protein-protein interaction networks were used to identify AD-associated genes that interact with multiple other genes, as drugs targeting them have large efficacies via co-regulation or interactions. RESULTS: Gene-based PRS identified 389 genes with 164 of them not previously reported as AD-associated. These 389 genes explained 56.12% -97.46% SNP heritability; and they were enriched in brain tissues and 164 biological processes, most of which are related to AD and other neurodegenerative diseases. We prioritized 688 drugs targeting 64 genes that were in the same co-expression modules and/or PPI networks. CONCLUSIONS: Gene-based PRS is a cost-effective way to identify AD-associated genes without substantially increasing the sample size. Co-expression modules and PPI networks can be used to identify drugs having large efficacies.

The collaborative study on the genetics of alcoholism: Genetics.

This review describes the genetic approaches and results from the family-based Collaborative Study on the Genetics of Alcoholism (COGA). COGA was designed during the linkage era to identify genes affecting the risk for alcohol use disorder (AUD) and related problems, and was among the first AUD-focused studies to subsequently adopt a genome-wide association (GWAS) approach. COGA's family-based structure, multimodal assessment with gold-standard clinical and neurophysiological data, and the availability of prospective longitudinal phenotyping continues to provide insights into the etiology of AUD and related disorders. These include investigations of genetic risk and trajectories of substance use and use disorders, phenome-wide association studies of loci of interest, and investigations of pleiotropy, social genomics, genetic nurture, and within-family comparisons. COGA is one of the few AUD genetics projects that includes a substantial number of participants of African ancestry. The sharing of data and biospecimens has been a cornerstone of the COGA project, and COGA is a key contributor to large-scale GWAS consortia. COGA's wealth of publicly available genetic and extensive phenotyping data continues to provide a unique and adaptable resource for our understanding of the genetic etiology of AUD and related traits.

Genome-wide case-only analysis of gene-gene interactions with known Parkinson's disease risk variants reveals link between LRRK2 and SYT10.

The effects of one genetic factor upon Parkinson's disease (PD) risk may be modified by other genetic factors. Such gene-gene interaction (G×G) could explain some of the 'missing heritability' of PD and the reduced penetrance of known PD risk variants. Using the largest single nucleotide polymorphism (SNP) genotype data set currently available for PD (18,688 patients), provided by the International Parkinson's Disease Genomics Consortium, we studied G×G with a case-only (CO) design. To this end, we paired each of 90 SNPs previously reported to be associated with PD with one of 7.8 million quality-controlled SNPs from a genome-wide panel. Support of any putative G×G interactions found was sought by the analysis of independent genotype-phenotype and experimental data. A total of 116 significant pairwise SNP genotype associations were identified in PD cases, pointing towards G×G. The most prominent associations involved a region on chromosome 12q containing SNP rs76904798, which is a non-coding variant of the LRRK2 gene. It yielded the lowest interaction p-value overall with SNP rs1007709 in the promoter region of the SYT10 gene (interaction OR = 1.80, 95% CI: 1.65-1.95, p = 2.7 × 10-43). SNPs around SYT10 were also associated with the age-at-onset of PD in an independent cohort of carriers of LRRK2 mutation p.G2019S. Moreover, SYT10 gene expression during neuronal development was found to differ between cells from affected and non-affected p.G2019S carriers. G×G interaction on PD risk, involving the LRRK2 and SYT10 gene regions, is biologically plausible owing to the known link between PD and LRRK2, its involvement in neural plasticity, and the contribution of SYT10 to the exocytosis of secretory vesicles in neurons.

Genomic risk for post-traumatic stress disorder in families densely affected with alcohol use disorders.

Recent genome-wide association studies (GWAS) have identified genetic markers of post-traumatic stress disorder (PTSD) in civilian and military populations. However, studies have yet to examine the genetics of PTSD while factoring in risk for alcohol dependence, which commonly co-occur. We examined genome-wide associations for DSM-IV PTSD among 4,978 trauma-exposed participants (31% with alcohol dependence, 50% female, 30% African ancestry) from the Collaborative Study on the Genetics of Alcoholism (COGA). We also examined associations of polygenic risk scores (PRS) derived from the Psychiatric Genomics Consortium (PGC)-PTSD Freeze 2 (N = 3533) and Million Veterans Program GWAS of PTSD (N = 5200) with PTSD and substance dependence in COGA, and moderating effects of sex and alcohol dependence. 7.3% of COGA participants met criteria for PTSD, with higher rates in females (10.1%) and those with alcohol dependence (12.3%). No independent loci met genome-wide significance in the PTSD meta-analysis of European (EA) and African ancestry (AA) participants. The PGC-PTSD PRS was associated with increased risk for PTSD (B = 0.126, p < 0.001), alcohol dependence (B = 0.231, p < 0.001), and cocaine dependence (B = 0.086, p < 0.01) in EA individuals. A significant interaction was observed, such that EA individuals with alcohol dependence and higher polygenic risk for PTSD were more likely to have PTSD (B = 0.090, p < 0.01) than those without alcohol dependence. These results further support the importance of examining substance dependence, specifically alcohol dependence, and PTSD together when investigating genetic influence on these disorders.

Clinical, genomic, and neurophysiological correlates of lifetime suicide attempts among individuals with alcohol dependence.

Research has identified clinical, genomic, and neurophysiological markers associated with suicide attempts (SA) among individuals with psychiatric illness. However, there is limited research among those with an alcohol use disorder, despite their disproportionately higher rates of SA. We examined lifetime SA in 4,068 individuals with DSM-IV alcohol dependence from the Collaborative Study on the Genetics of Alcoholism (23% lifetime suicide attempt; 53% female; 17% Admixed African American ancestries; mean age: 38). We 1) explored clinical risk factors associated with SA, 2) conducted a genome-wide association study of SA, 3) examined whether individuals with a SA had elevated polygenic scores for comorbid psychiatric conditions (e.g., alcohol use disorders, lifetime suicide attempt, and depression), and 4) explored differences in electroencephalogram neural functional connectivity between those with and without a SA. One gene-based finding emerged, RFX3 (Regulatory Factor X, located on 9p24.2) which had supporting evidence in prior research of SA among individuals with major depression. Only the polygenic score for suicide attempts was associated with reporting a suicide attempt (OR = 1.20, 95% CI = 1.06, 1.37). Lastly, we observed decreased right hemispheric frontal-parietal theta and decreased interhemispheric temporal-parietal alpha electroencephalogram resting-state coherences among those participants who reported a SA relative to those who did not, but differences were small. Overall, individuals with alcohol dependence who report SA appear to experience a variety of severe comorbidities and elevated polygenic risk for SA. Our results demonstrate the need to further investigate suicide attempts in the presence of substance use disorders.

Unsupervised representation learning improves genomic discovery and risk prediction for respiratory and circulatory functions and diseases.

High-dimensional clinical data are becoming more accessible in biobank-scale datasets. However, effectively utilizing high-dimensional clinical data for genetic discovery remains challenging. Here we introduce a general deep learning-based framework, REpresentation learning for Genetic discovery on Low-dimensional Embeddings (REGLE), for discovering associations between genetic variants and high-dimensional clinical data. REGLE uses convolutional variational autoencoders to compute a non-linear, low-dimensional, disentangled embedding of the data with highly heritable individual components. REGLE can incorporate expert-defined or clinical features and provides a framework to create accurate disease-specific polygenic risk scores (PRS) in datasets which have minimal expert phenotyping. We apply REGLE to both respiratory and circulatory systems: spirograms which measure lung function and photoplethysmograms (PPG) which measure blood volume changes. Genome-wide association studies on REGLE embeddings identify more genome-wide significant loci than existing methods and replicate known loci for both spirograms and PPG, demonstrating the generality of the framework. Furthermore, these embeddings are associated with overall survival. Finally, we construct a set of PRSs that improve predictive performance of asthma, chronic obstructive pulmonary disease, hypertension, and systolic blood pressure in multiple biobanks. Thus, REGLE embeddings can quantify clinically relevant features that are not currently captured in a standardized or automated way.